NLP researcher working on efficiency benchmarks for tool-using LLM agents; first author of RideWay (2026).
1
papers here
1
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-20) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Not yet scored. This axis measures total research output — publication count, venues, and authorship roles. The catalog holds only one 2026 arXiv preprint, too little to estimate productivity, so no figure is given.
Impact—
Not yet scored. Impact would capture citations, uptake by other benchmarks, and influence on agent evaluation practice. With a single 2026 preprint and no citation data in scope, any estimate would be invented rather than measured.
Focus100/100
Focus scores 100: every catalogued output sits in one field, natural language processing / LLM agent evaluation (one paper). The profile is sharply defined around agent benchmarking and efficiency measurement, though with a single reviewed paper this concentration partly reflects sample size as much as deliberate specialization.
Collaboration28/100
Collaboration scores 28, computed from three distinct coauthors — Boli Fang, Mingzhi Hou, and Claire Liu — all on the same first-authored preprint. This points to a small, stable working group rather than a broad network; no cross-institution pattern can be confirmed from the current record.
Activity—
Not yet scored. Activity tracks publishing cadence — papers per year and continuity over time. Only one 2026 entry exists, so no rhythm or sustained-output signal can be established yet.
Reproducibility100/100
Reproducibility scores 100 — a perfect mean computed across one assessed paper. RideWay is rated high on reproducibility readiness, and Evolve's review describes the work as rigorous and candid about its own limitations, consistent with transparent methodology. The evidence base remains a single paper.
Research summary
Qingnuan Han is an early-stage natural language processing researcher whose visible record centers on evaluation methodology for large language model (LLM) agents. Evolve's reviewed catalog contains one paper: the 2026 arXiv preprint RideWay: Benchmarking Efficient Task Completion for Tool-Using Language Agents, with Han as first author. The paper's core move is to add a human-calibrated efficiency metric on top of agent task success, instantiated in a ridehailing setting and calibrated with a Bradley-Terry model. Evolve's review calls the contribution rigorous and candid while flagging two weaknesses — the tool axis performs at chance level and labels are thin. Bibliometric signals are narrow but internally consistent: a focus score of 100 reflects output confined to a single field (natural language processing / LLM agent evaluation); a reproducibility score of 100 derives from one assessed paper rated high on reproducibility readiness; and a collaboration score of 28 comes from three distinct coauthors (Boli Fang, Mingzhi Hou, Claire Liu). Output, impact, and activity dimensions are not yet scored because the underlying data is insufficient. This is a full-tier but early profile: substantive on what exists, silent on what does not.
Collaboration pattern
Han's visible collaboration footprint is compact: three distinct coauthors — Boli Fang, Mingzhi Hou, and Claire Liu — appear alongside a first-authored 2026 preprint, yielding a collaboration score of 28. The pattern suggests a small, stable working group rather than a wide network, consistent with an early-stage profile. No cross-institution or senior-collaborator structure can be inferred from the available data, since institutional affiliations are not part of the current record.
Research keywords
Highlights
Human-calibrated efficiency metric for tool-using agents
Perfect reproducibility score (100) from a high-readiness assessment
Candid limitations: chance-level tool axis, thin labels
Timeline
- 2026First-authored arXiv preprint RideWay appears, benchmarking efficient task completion for tool-using language agents.
- 2026Evolve reviews RideWay, judging it rigorous and candid while flagging a chance-level tool axis and thin labels.
- 2026Reproducibility assessment rates the paper high, contributing a reproducibility mean of 100 to the profile.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Dimension scores are derived from open bibliometric data (OpenAlex-based snapshots); citation and venue coverage may be incomplete.
- Only one reviewed paper — a 2026 arXiv preprint — is in scope, so the visible record likely understates total output.
- No institution, country, or ORCID is available; identity is name-resolved from public bibliographic data, and 'Qingnuan Han' may be shared by other researchers.
- Output, impact, and activity dimensions are unscored due to insufficient data; no citation counts are quoted anywhere in this profile.